Tally 7.1
Before: You had to log in to your GST account every day, and visit the GST portal separately just to validate HSN/SAC codes.
Now: Tally 7.1 fixes both. Stay logged in to your GST account for up to 30 days — no more daily logins — and validate HSN/SAC codes right inside Tally, straight from your masters and reports like GSTR-1 and the HSN/SAC Summary, with no separate site visit needed.
Microsoft 365
Copilot Cowork Goes GA. Copilot Cowork is now generally available, bringing enterprise-ready AI for long-running, multi-step work into Microsoft 365 Copilot. It automates complex tasks across your business data, systems, and connected apps with grounded context, multi-model intelligence, and integrated plugins.
It runs on the same security and governance controls you already trust — so teams can hand off longer-running work without stepping outside familiar compliance boundaries.
Perplexity
Claude Opus 5, Live and Cheaper. Claude Opus 5 is now live across Perplexity Search and Computer. On the WANDR benchmark it outperformed every model tested except Fable 5, while costing 57% less per task than the previous flagship.
That means frontier-level reasoning and agentic capability for demanding tasks, longer Computer workflows at a lower cost per task, and a clear way to weigh cost against quality using the WANDR benchmark results.
AI-Operator Malware. Hackers are bypassing basic prompt guardrails by planting infected configuration files that AI agents automatically load and trust. Using this method, a single developer deployed an 88,000-line rogue command-and-control framework called VoidLink in under a week.
The Vulnerability Flood. Highlighting the sheer volume of software defects, Microsoft's July Patch Tuesday broke all historic records by addressing a staggering 622 unique vulnerabilities across its business systems — a surge driven by AI-powered multi-model scanning engines discovering bugs at machine speed.
Exploits Built in 24 Hours. Security reports indicate attackers using AI have drastically cut the time needed to build a working software exploit — from 72 hours down to just 24 — even as AI tools help vendors patch flaws faster internally.